Limits HTML to a certain number of words. Is aware of tags etc and will not count them in the word-count, as well as closing them properly. This doesn't actually pass all unit tests at the moment - an exact match in num words will still put in ... part.